CS5228 Assignment Brief

In the era of data-driven decision making, knowledge discovery and data mining (KDDM) play an essential role in transforming raw data into meaningful insights. One widely adopted framework is CRISP-DM, which provides structured steps for carrying out real-world data mining projects across industries.

This assignment allows you to take on the role of a data analyst or data scientist solving a real-world problem using textual data. Your objective is to apply the CRISP-DM methodology using a real-world text dataset and demonstrate your understanding of the complete text mining and knowledge discovery process. You are expected to deliver a well-documented, functional, and insightful analytical product.

The CRISP-DM process includes:

1. Business Understanding
2. Data Understanding
3. Text Data Preparation
4. Modeling (at least two models)
5. Evaluation
6. Deployment (suggested application)

Each phase should be clearly addressed in your project report, with appropriate justifications, visuals, and insights. Emphasis should be placed on transparency, reproducibility, and the relevance of your analysis to the chosen business or application context.

You may source datasets from reliable open data repositories such as Kaggle, UCI Machine Learning Repository, data.gov.my, or other publicly accessible text-based datasets. Ensure the data you select has enough depth and variety to support meaningful analysis.

You are free to choose any domain (e.g., healthcare, retail, social media, finance, environmental science), as long as:

1. The dataset is relevant, sufficient, and manageable
2. The problem statement is well defined
3. The solution demonstrates the application of a text mining or data mining techniques (such as text classification, sentiment analysis, clustering, topic modeling, or spam detection)

Creativity, technical rigor, and clear presentation of findings will be key to achieving a high score. Ethical considerations (e.g., bias handling and responsible use of textual data) are encouraged and rewarded where appropriate.

Requirements

1. If you do not attend the walkthrough the maximum mark you can achieve for this assignment is 40%.
2. Please do not submit hand-drawn diagrams. Hand-drawn diagrams or hand-written reports will receive zero (0) marks.
3. Your submission documentation’s content should include the following items:
   • Report
   • Assessment rubric

Assessment Critera

Report: 20%

1. Business Understanding – 10% of marks

Clearly define the business problem or application domain addressed in the project. Explain the project objectives, expected outcomes, stakeholders involved, and the relevance of the selected text dataset. Justify why the problem is important and how text mining can contribute to solving it.

2. Data Understanding – 15% of marks

Describe the selected dataset, including its source, size, attributes, and characteristics. Perform exploratory data analysis (EDA) using appropriate statistics and visualizations. Identify data quality issues, class distribution, potential challenges, and key insights obtained from the textual data.

3. Text Data Preparation – 20% of marks

Provide a complete description of all preprocessing activities performed on the text data. This may include data cleaning, tokenization, stop-word removal, stemming, lemmatization, vectorization (e.g., TF-IDF, Count Vectorizer), feature engineering, and dataset splitting. Justify the techniques selected and explain their impact on the analysis.

4. Modeling – 25% of marks

Develop and implement at least two text mining or machine learning models. Clearly describe the algorithms used, model configurations, parameter settings, and training procedures. Justify the selection of models and explain how they address the problem statement.

5. Evaluation – 20% of marks

Evaluate and compare the performance of the developed models using appropriate metrics such as Accuracy, Precision, Recall, F1-Score, Confusion Matrix, ROC-AUC, or other relevant measures. Discuss findings, strengths, limitations, and provide insights into model performance.

6. Deployment (Suggested Application) – 10% of marks

Propose a practical deployment scenario for the developed solution. Explain how the model could be integrated into a real-world application, system, or business process. Include a conceptual architecture, prototype, dashboard, web application, or workflow diagram where appropriate.